Download Does Annual Real Gross Domestic Product per Capita Overstate or

SUBSCRIBE NOW AND RECEIVE
CRISIS AND LEVIATHAN* FREE!
“The Independent Review does not accept
pronouncements of government officials nor the
conventional wisdom at face value.”
—JOHN R. MACARTHUR, Publisher, Harper’s
“The Independent Review is
excellent.”
—GARY BECKER, Noble Laureate
in Economic Sciences
Subscribe to The Independent Review and receive a
free book of your choice* such as the 25th Anniversary
Edition of Crisis and Leviathan: Critical Episodes in the
Growth of American Government, by Founding Editor
Robert Higgs. This quarterly journal, guided by co-editors
Christopher J. Coyne, and Michael C. Munger, and Robert
M. Whaples offers leading-edge insights on today’s most
critical issues in economics, healthcare, education, law,
history, political science, philosophy, and sociology.
Thought-provoking and educational, The Independent
Review is blazing the way toward informed debate!
Student? Educator? Journalist? Business or civic leader?
Engaged citizen? This journal is for YOU!
* Order today for more FREE book options
Perfect for students or anyone on
the go! The Independent Review is
available on mobile devices or tablets:
iOS devices, Amazon Kindle Fire, or
Android through Magzter.
INDEPENDENT INSTITUTE, 100 SWAN WAY, OAKLAND, CA 94621 • 800-927-8733 • [email protected] PROMO CODE IRA1703
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 181
Does Annual Real Gross
Domestic Product per Capita
Overstate or Understate the
Growth of Individual
Welfare over the Past Two
Centuries?
—————— ✦ ——————
HENDRIK VAN DEN BERG
The decrease in the death rate, and the attendant increase in life
expectancy—more than doubling—during the last two centuries in the
richer countries, and in the twentieth century in the poorer countries, is
sic the most stupendous feat in human history.
—Julian Simon, introduction to The State of Humanity (26)
Angus Maddison’s (2001) historical estimates of annual real gross domestic product (GDP) per capita show that nearly all the economic growth in annual real
GDP per capita that has ever occurred in the world occurred after 1820. Maddison’s estimate of 1820 real GDP per capita of $667 at 1990 prices and others’
estimates that the real output per capita at which humanity can survive is approximately $250 per year at 1990 prices imply that average world real GDP per capita
little more than doubled during the first several hundred thousand years of
Hendrik Van den Berg is an associate professor in the Department of Economics at the University of
Nebraska, Lincoln.
The Independent Review, v.VII, n.2, Fall 2002, ISSN 1086-1653, Copyright © 2002, pp. 181– 196.
181
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 182
182
✦
H E N D R I K VA N D E N B E R G
human existence.1 During the past two hundred years, the world’s average real
GDP per capita has increased nearly ninefold. In the more developed countries,
the growth of annual GDP per capita has been much greater. Figure 1 illustrates
the recent explosion of the world’s average annual real GDP per capita.
The accuracy of the stunning picture of recent economic growth shown in figure
1 depends on whether annual real GDP per capita correctly reflects human welfare and
economic growth. Many economists have accepted real GDP per capita as a reasonable
proxy for human welfare. For example, Simon Kuznets explicitly defined economic
growth as “a sustained increase in per capita or per worker product” (1966, 1), and
Douglass C. North and Robert Thomas wrote that economic growth “implies that the
total income of society must increase more rapidly than population” (1973, 1). However, although few people deny that Maddison’s carefully researched historical data indicate what happened to annual real GDP per capita over the past two hundred years,
some authors have objected to using real GDP per capita to measure human welfare.
A common theme in the economic development literature has been that the
growth in average annual real GDP per capita overstates the improvement in human
welfare.2 Some economists have suggested alternative measures of economic growth
Figure 1
Annual Real GDP per Capital: 0–1998
Real GDP Per Capita
$5,000
$4,000
$3,000
$2,000
$1,000
0
1000
1500
1820 1900 1998
Source: Based on data from Maddison 2001.
1. See Pritchett 1997 for a justification of the $250 subsistence-income estimate.
2. See, for example, the discussions of the well-known Human Development Index (HDI) published annually by the United Nations Development Program (UNDP). See also Usher 1980 and Meier 1976, 12–18.
GDP-based growth measures continue to receive criticism; see, for example, Briscoe 2001. Van den Berg
2001 describes the many alternative measures of economic growth.
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 183
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
183
that focus on health, education, political rights, and many other conditions that are
equated with the quality of life. Julian Simon (1995a) and others have pointed out that
virtually all measures of human welfare have improved greatly over the past two hundred years, in which case the general impression given by figure 1 is not misleading.3
However, because most of the other measures have not improved by as great a multiple as real GDP per capita, the perception remains that the growth in annual GDP per
capita overstates the improvements in human welfare during the past two centuries.
In this article, I argue that the various alternative measures of human welfare do
not remedy a serious fundamental shortcoming of annual real GDP per capita as an
indicator of human welfare, which is that it fails to measure the lifetime welfare of
individuals. The many alternative measures of human welfare that have been proposed are, like GDP, also annual flows or point-in-time stocks, none of which tells us
anything about individuals’ lifetime welfare. A simple measure of average individual
lifetime welfare fortunately can be constructed easily from readily available data. This
measure indicates that the recent growth of individual welfare actually has exceeded
the growth indicated by annual real GDP per capita.
Alternative Measures of Human Welfare
Economists have suggested many alternative measures of human welfare, including
life expectancy, infant mortality rate, caloric intake, access to safe water, adult literacy
rate, school enrollment, the distribution of income, hospital beds per capita, and a
long list of other similar measures of the “quality of life.” One especially popular
approach to improving on GDP measures of growth has been to use a weighted average of several alternative measures of human welfare. Typical of this approach is the
well-known Human Development Index (HDI) published annually by the United
Nations Development Program (UNDP). The HDI is a weighted average of real
GDP per capita, a measure of health, and a measure of education.
Representing many development economists’ belief that the welfare effect of marginal income declines as income per capita rises, the HDI decreases the weight of real
GDP per capita relative to the other measures as real output per capita rises. In fact,
until recently, increases in real output per capita carried weight in the HDI only for low
levels of annual output per capita.4 Specifically, before 1999, the relationship between
annual real GDP per capita and human welfare that the UNDP assumed was the one
depicted by the solid “kinked” line in figure 2. Thus, increases in GDP per capita up to
the world average of about $5,000 had a strong effect on the value of the overall HDI
index because, as the solid line in figure 2 shows, in that range small increases in GDP
per capita were assumed to bring large increases in human welfare. After the world’s
average GDP per capita of $5,000 was reached, however, further increases in output
3. See, for example, the many papers included in Simon 1995b.
4. The methods used to calculate the HDI are described in the “Technical Note” in the UNDP 1999,
159–60.
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 184
184
✦
H E N D R I K VA N D E N B E R G
Figure 2
Welfare and GDP: UNDP
Welfare
1.000
0
$5,000
$10,000
$15,000
$20,000
$25,000
GDP
were assumed to matter only slightly for human welfare, and hence further growth of
output per capita was deemed to be of little importance. This approach meant that the
great differences in real output per capita between Japan and a middle-income country
such as Mexico made almost no difference for the two countries’ relative HDI indices.
The idea behind this approach was that people have certain “basic needs,” and once
these needs are satisfied, further increases in consumption will be largely superfluous.5
Beginning with the 1999 HDI, a more “curved” relationship between income
and welfare was assumed, at the urging of Sudhir Anand and Amartya Sen (1999). Sen
writes: “we have excellent reasons for wanting more income or wealth. This is not
because income and wealth are desirable for their own sake, but because, typically, they
are admirable general-purpose means for having more freedom to lead the kind of lives
we have reason to value. . . . The usefulness of wealth lies in the things that it allows us
to do—the substantive freedoms it helps us to achieve” (1999, 14). In other words,
economic growth increases positive economic freedom, and that increase has real welfare implications. As a result of such reasoning, the 1999 HDI was modified so that the
5. Ideology may also have inspired the UNDP’s procedure: the UNDP has often emphasized distribution
over absolute growth in the belief that people with “below average” incomes are justified in aspiring to
higher incomes, but those with high incomes engage in only superfluous consumption.
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 185
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
185
relationship between real GDP per capita and individual welfare was assumed to be the
one shown by the dashed curve in figure 2 (UNDP 1999). Note, however, that
although the revision of the HDI accepts that welfare always is related positively to real
GDP per capita, the HDI still does not assume a constant one-to-one relationship.
Apart from the possibility that welfare gains from GDP growth might be nonlinear, the growth of annual real GDP per capita may exaggerate improvements in
human welfare for other reasons. A well-known shortcoming of GDP is that it counts
only legal market transactions, omitting household production, illegal or “underground” production, and leisure despite their effects on human welfare. The omission
of household and underground activity, for example, causes the growth in official
GDP to overstate true growth in production if the unrecorded economic activity
shrinks as a percentage of total economic activity. As Scott Fuess and I (Fuess and Van
den Berg 1996, 1998) found for the United States and Mexico, this development
probably has occurred in many economies as women increasingly have entered the
formal labor market and have substituted paid work for previously unvalued but valuable housework. GDP may also overstate real growth in output because it does not
account for the depletion of nonrenewable resources. Martin Weitzman (1999) has
estimated the cost of resource depletion in order to derive net domestic product
(NDP), and he has found that the annual depletion of the world’s major minerals is
equivalent to approximately 1 percent of real GDP per year.
Growth of Real GDP per Capita May Understate
Improvements in Human Welfare
Not all evidence supports the belief that the estimated growth of annual real GDP per
capita overstates the growth of human welfare. Michael Boskin and others (1996)
contend that the U.S. government’s consumer price index consistently has overstated
annual inflation by approximately one percentage point because it has not accounted
adequately for quality improvements or product substitution.6 If inflation has been
overstated, then the growth rate of annual real GDP per capita has been understated
by one percentage point per year, and the growth of material wealth since 1900 has
been more than double what GDP per capita indicates.
Annual real GDP per capita also ignores the fact that economic growth usually
increases not just the quantity of output but also the variety of output. Increased variety, all other things being equal, enhances welfare (Federal Reserve Bank of Dallas
1999). Because GDP is a measure of aggregate output, it makes no difference for
GDP whether you eat the same $5.75 lunch consisting of a hamburger, french fries,
and a Coca-Cola every day of the week or whether you eat the $5.75 hamburger dish
only on Monday, and you then eat a $4.50 Mexican taco platter on Tuesday, a $4.00
6. The findings of Boskin and others 1996 are summarized conveniently in Boskin and others 1998.
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 186
186
✦
H E N D R I K VA N D E N B E R G
spaghetti dish on Wednesday, a $5.00 Chinese meal on Thursday, and a $9.50 Japanese sushi lunch on Friday. The value of the five lunches is $28.75 in either case, and
GDP will be the same in either case. For most people, however, the variety of five different meals is preferable to eating the same hamburger platter every day of the week.
The preceding discussion suggests that we are not yet close to a consensus on
whether annual real GDP per capita overstates the improvement in human welfare over
the past two centuries. The discussion, however, does bring out a more subtle and more
important point, which is that economists and statisticians have missed a critical weakness of annual real GDP per capita and of all the alternative measures that have been proposed, despite all their efforts to improve the measurement of economic growth.
Annual output or any other measure used to calculate welfare at some point in time cannot capture welfare as individuals experience it over the course of a lifetime.
How Should We Measure
the Growth of Individual Welfare?
Real GDP per capita is an aggregate value measured over some arbitrary brief period
and divided by the number of people alive at that time. Individuals live for more than
one year, however, and their personal welfare depends on a lifetime’s consumption of
goods and services. In this regard, none of the popular alternative proxies for human
welfare is any more appropriate than real GDP per capita. For example, measures such
as literacy, years of schooling, political freedom, clean air, or the number of hospital
beds are also either annual flows or point-in-time quantities, not lifetime totals.
It is easy to show that the current practice of using annual real GDP per capita
to measure human welfare gives us a hopelessly inaccurate measurement of individual
welfare. Suppose that two countries have the identical annual real GDP per capita of
$10,000. Suppose, also, that in one country everyone lives to the age of eighty, but in
the second country people live only to the age of thirty. In which country are individuals better off? To suppose that these two countries provide their residents with the
same levels of welfare would be unreasonable. The human welfare that interests economists cannot be captured by a collective measure of average real GDP or by the average of any of the popular alternative measures of welfare, quantified over some brief
period or at some point in time.
Individuals judge their welfare by the quality of life and by the length of time that
they can experience a particular quality of life. Therefore, a measure of an economy’s
capacity to satisfy human wants must take into account not only how much people have
at some moment or period of time, but also how long the economy can keep people
alive and able to experience the annual flows of goods and services that the economy
produces. An acceptable measure of human welfare must be a lifetime measure.
A better measure of individual human welfare would be the average annual GDP
per capita times the average life expectancy that the economy provides its citizens. The
same point applies to the many alternative welfare measures based on point-in-time
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 187
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
187
quantities or on annual flows. If education is deemed to be a better measure of human
welfare, the effect of education on human welfare depends on how long people live to
benefit from their education. If health or economic freedom is an important yardstick
with which to measure human welfare, we must adjust these measures for the length
of a lifetime that the average person experiences health or economic freedom.
If an economy provides the conditions under which individuals can live longer, that
provision in itself increases individual welfare. If the economy can both keep people alive
longer and provide more goods, better health, and more education, then the welfare of
individuals is enhanced further. Some of the different measures of economic growth—
such as life expectancy, real income per capita, or even weighted averages of variables that
reflect quality of life—are in many cases not alternatives but multiplicatives.
Expected Individual Lifetime Welfare
The most direct approach to deriving a measure of lifetime individual welfare, for
which data are readily available, is to calculate the product of annual real GDP per
capita and life expectancy at birth. Life expectancy in a given year is the average age to
which a group of newborns would live if the members of that group remained subject
to the age-specific death rates prevailing in their year of birth. Annual real GDP per
capita, of course, indicates the value of goods and services that the economy is capable of providing the average person in a given year. Hence, the product of the two
measures indicates the economy’s capacity in a certain year (1) to provide material
welfare and (2) to sustain an individual life. Frank Lichtenberg (1998) introduced just
such a measure, which he denotes as YL (YA)(E), where YL is expected lifetime GDP
per capita, YA is average annual GDP per capita, and E is life expectancy. Lichtenberg
introduced this measure as part of his study to estimate the value of pharmaceutical
innovation.7 This measure can be used for studies of economic growth, human welfare, and overall economic performance because the data needed to calculate it are
readily available. In the remainder of this article, I refer to this multiplicative measure
as the expected individual lifetime welfare (EILW).
Using Maddison’s (2001) data, the Indian economy in 1998 had the capacity to
provide the average person with $1,746 worth of goods and services each year and to
keep the person alive to enjoy that amount of consumption for sixty years. Thus,
Indian EILW was equal to $104,760. In the United States in 1998, the economy provided on average $27,331 worth of goods and services under conditions that would
let people live, on average, for seventy-seven years. Thus, U.S. EILW in 1998 was
$2,104,487. Although the U.S. real GDP per capita is a little less than sixteen times
as great as that of India, its measure of expected lifetime welfare per capita is more
than twenty times as great. The average American gets to enjoy the (higher) income
7. Van den Berg 2001 calculates a historical series of Lichtenberg’s measure for India, Mexico, Spain, Russia, and the United States, based on data from Maddison 1995.
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 188
188
✦
H E N D R I K VA N D E N B E R G
seventeen years longer than the average Indian gets to enjoy the (lower) income.
Thus, the average American is relatively better off than either GDP per capita or life
expectancy by itself indicates.
Lifetime Welfare Should Not Be Discounted
Almost thirty years ago, Dan Usher (1973) suggested the use of discounted lifetime
earnings as a measure of individual welfare. More recently, William Nordhaus (2002)
and Kevin Murphy and Robert Topel (1999) have used the discounted flow of lifetime consumption to measure the gains in human welfare from improved health.8 We
are indebted to Usher for introducing the idea of using a lifetime measure as a proxy
for national welfare, but his discounting procedure is not appropriate for measuring
the economy’s capacity to provide a lifetime of welfare-enhancing output.9 It is worth
a brief digression to point out why discounting is not appropriate, given economists’
natural tendency to discount lifetime flows of income.
By discounting the future flow of income, present income would be valued
more highly than income that arrives later in life. There is no obvious reason to
weight one year more highly than any other in calculating lifetime welfare.10 There
is substantial evidence that at any single point in time, people value the present
most highly and discount both the future and the past. This discounting of both the
present and the past implies a well-known time inconsistency problem: later in life
people are likely to regret what they did earlier in life.11 Older people will wish they
had studied harder, saved more, drunk less, and so forth when they were young.
The importance of this observation is that in calculating lifetime welfare it is necessary to measure a person’s evaluation of welfare at each period of a person’s life, not
from the perspective of one point in time as in the case of discounted lifetime
income per capita.
Figure 3 illustrates a series of hypothetical discounted views of lifetime income at
different ages in an individual’s life. Individuals discount both the past and the future.
They of course may discount the past differently than they discount the future, and
they may discount the past and future differently at different ages. The curves may
not be symmetrical in both directions from their peaks. The peaks may differ in height
for different years of individuals’ lives; for example, children may view a given level of
8. The papers by Lichtenberg (1998), Murphy and Topel (1999), and Nordhaus (2002) will be published
by the University of Chicago Press in a forthcoming volume edited by Murphy and Topel. Murphy and
Topel’s research is summarized in Murphy and Topel 2000.
9. Usher (1980) uses a similar approach.
10. Carryover from the welfare of previous years or anticipation of the welfare to be experienced in future
years may augment or diminish the welfare experienced in any given year, but that possibility does not imply
that future welfare necessarily must be discounted and converted to a present welfare value.
11. For a discussion of bidirectional discounting and the time inconsistency problem that it creates, see
Caplin and Leahy 2000.
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 189
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
189
Figure 3
Discounting the Future and the Past
Discount
1
0
t=1
t=2
t=3
t=4
t=5
t=6
material welfare differently than adults. Hence, the shapes of the curves in figure 3
should be seen as illustrative only. What is clear, however, is that lifetime welfare is the
sum of the welfare values enjoyed in each year or at each point in time, which is shown
by the peaks of the curves. As an approximation, it is clearly more accurate to treat
lifetime income as a sum of equal peaks, given by average annual real income per
capita, as Lichtenberg suggests, than as a series of continually decreasing discounted
future income flows, as Usher has proposed.
Growth of Expected Individual
Lifetime Welfare Since 1820
Table 1 presents measures of EILW for a group of countries for which both real GDP
per capita and life expectancy data are available from 1820 to the present. The table
makes it clear that our understanding of economic growth changes substantially when
we measure it by EILW. The EILW data specifically show that the growth of human
welfare has been much greater than the rate of growth calculated using just real GDP
per capita. In all countries, EILW growth was faster than the growth of real GDP per
capita. In China, for example, EILW growth was positive between 1900 and 1950
even though real GDP per capita shrank. In India, EILW grew nearly twice as fast as
real GDP per capita. Only in the case of Russia from 1950 to 1998 did EILW growth
differ little from the growth of real GDP per capita—Russia and its predecessor, the
USSR, experienced virtually no improvement in life expectancy over this period.
VOLUME VII, NUMBER 2, FALL 2002
THE INDEPENDENT REVIEW
4,593
4,096
Sweden
United Kingdom 1,707
United States
740
760
600
531
667
Mexico
China
India
World
689
Russia
Brazil
669
Japan
1,257
2,561
1,063
1,198
Spain
1,746
1,263
625
540
1,157
704
1,023
1,135
2,040
3,533
1,117
1,821
3,134
2,849
Netherlands
1,058
Italy
1,230
Germany
2,114
619
439
2,365
1,672
2,834
1,926
9,561
6,907
6,738
2,397
5,996
3,502
3,881
5,270
5,709
1,746
3,117
6,655
5,459
3,893
20,413
27,331
18,714
18,685
14,227
20,224
17,759
17,779
19,558
1998
26
21
—
—
27
28
34
39
40
39
28
32
30
41
37
31
24
24
33
36
32
44
47
50
56
35
52
43
47
47
1900
49
32
41
50
45
65
61
68
69
70
62
72
66
67
65
1950
66
60
71
72
67
67
81
77
77
79
78
78
78
77
78
1998
1820
1950
1820
1900
Years of Life Expectancy
at Birth
______________________________
Annual Real GDP per Capita
(in dollars)
_____________________________
17,342
11,151
—
—
19,980
19,292
22,746
49,023
68,280
46,772
29,764
58,272
33,510
43,378
45,510
1820
39,153
15,000
12,960
38,181
25,344
32,736
49,940
192,512
229,650
143,416
71,400
183,716
75,078
147,298
133,903
1900
103,586
19,808
17,999
118,250
75,260
184,210
117,486
650,148
476,583
471,660
148,614
431,712
231,132
260,027
342,550
1950
376,794
104,760
221,307
479,160
365,753
260,831
1,653,453
2,104,487
1,440,978
1,476,115
1,109,706
1,577,472
1,385,202
1,370,523
1,525,524
1998
EILW
(in
dollars)
______________________________________
✦
France
Levels:
190
Expected Lifetime Welfare Per Capita: Levels and Annual Growth Rates for 1820–1998
Table 1
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 190
H E N D R I K VA N D E N B E R G
0.13
0.20
0.80
China
India
World
1.04
0.00
0.41
1.44
1.75
2.05
1.06
1.71
0.82
1.95
0.32
1.06
1.40
0.43
1.24%
1900-50
2.09
2.18
4.17
2.18
2.50
0.66
5.04
2.21
2.10
2.15
3.78
2.57
3.44
3.22
2.77%
1950-98
0.22
0.17
—
—
0.36
0.17
0.32
0.23
0.28
0.45
0.28
0.61
0.45
0.17
0.30%
1820-1900
0.92
0.58
1.08
0.83
0.45
1.42
0.66
0.74
0.65
0.45
1.15
0.65
0.86
0.71
0.65%
1900-50
0.61
1.29
1.13
0.76
0.82
0.06
0.58
0.25
0.22
0.25
0.48
0.16
0.34
0.28
0.37%
1950-98
Years of Life Expectancy
at Birth
____________________________
1.02
0.37
—
—
0.30
0.66
0.99
1.72
1.53
1.41
1.10
1.45
1.01
1.54
1.36%
1820-1900
1.96
0.56
0.66
2.29
2.20
3.52
1.73
2.46
1.47
2.41
1.48
1.72
2.31
1.14
2.17%
1900-50
2.73
3.53
5.37
2.96
3.35
0.73
5.66
2.48
2.33
2.41
4.28
2.74
3.80
3.52
1.90%
1950-98
EILW
______________________________
Sources: Maddison 2001 table 1–5a, table B-21. Because Maddison 2001 does not have real GDP per capita for 1900, I used Angus Maddison 1995,
table D, which, in most cases, corresponds perfectly or very closely to Maddison 2001; the few differences were not judged to be significant.
0.06
0.53
0.50
Russia
Mexico
0.66
Japan
Brazil
1.24
1.49
0.95
Sweden
United States
0.82
Spain
United Kingdom
0.56
0.83
Netherlands
Germany
Italy
1.06%
1.37
France
1820-1900
Annual Growth Rates:
Real GDP Per Capita
____________________________
Table 1 (continued)
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 191
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
191
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 192
192
✦
H E N D R I K VA N D E N B E R G
The more rapid growth of EILW means that the growth of annual real GDP per
capita understates the growth of lifetime material welfare. Even if the relationship
between output and human welfare is nonlinear, human welfare still might have
increased more rapidly than the growth of annual real GDP per capita indicates.
Convergence and Divergence
of National Welfare Levels
Because the EILW permits a more accurate depiction of individual welfare, it can also
help us get a better answer to the question of whether individual welfare is becoming
more or less equal among the countries of the world. The rate of absolute convergence of real income per capita across countries or regions can be found by estimating the regression equation
ln(yt+n) – ln(yt) a b ln(yt),
in which yt is real income per capita in year t. If the coefficient b is negative, we can
conclude that real incomes per capita converged over the period from year t to year t
n. The speed of convergence is equal to ß – ln(1 b)/n, and the often-reported
half-life, t*, of the absolute income gap is the solution to e–ßt* 0.5, or t* –
ln(0.5)/ß. Table 2 presents regression estimates and, where the regression coefficients were negative, the speed of convergence for both real GDP per capita and the
EILW for the countries included in table 1.
Table 2 makes it clear that average individual welfare across countries diverged
sharply during the 1800s. It makes little difference whether annual real GDP per
capita or the EILW is used to compare welfare across countries. The coefficients from
regressing subsequent growth on the initial levels of annual real GDP per capita and
the EILW are positive and nearly identical for the period 1820–1900.
In the twentieth century, there are substantial differences between what real GDP
per capita and the EILW show about the convergence of average individual welfare.
For the period 1900–1950, the regression coefficients for both annual real GDP per
capita and the EILW are still positive, but they are much smaller than for the nineteenth century. Note also that the coefficient for the EILW is considerably smaller than
in the regression equation relating the growth and initial levels of annual real GDP per
capita. Thus, individual welfare appears to have diverged less rapidly during the first
half of the twentieth century than data for annual real GDP per capita indicate.
During the second half of the twentieth century, the relatively more rapid convergence of the EILW, compared to annual real GDP per capita, indicates that individual welfare probably converged more rapidly than comparisons of real GDP per capita
suggest. The half-life of the differences in welfare across the sample of countries is
approximately only 100 years in the case of the EILW, but it is approximately 160 years
for annual real GDP per capita.
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 193
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
193
Table 2
Convergence of Human Welfare for Twelve Countries: 1820–1998
1820–1900
1900–1950
1950–1998
0.65800
0.00632
0.13160
0.00247
0.18756
0.00433
0.63230
0.00612
0.08720
0.00167
0.27650
0.00674
Real Per Capital GDP
Regression coefficient:
Speed of Convergence:
EILW:
Regression coefficient:
Speed of Convergence:
Some Qualifications and Extensions
The growth of EILW may still understate the gains in human welfare brought about
by increases in life expectancy. GDP does not account for leisure. Therefore, EILW,
which is simply annual real GDP per capita multiplied by life expectancy, also does
not take into account this other source of individual welfare. Given that work hours
tend to decline and leisure tends to increase as income per capita rises, the growth
of EILW will tend to understate the growth in actual individual welfare. Robinson
(1995), Nordhaus (2002), and Murphy and Topel (1999) effectively account for
leisure as well as for material wealth by using estimates of the “value of life” to arrive
at their measures of lifetime welfare for people currently living in the United
States.12 Their estimates of average lifetime welfare for the United States greatly
exceed the much simpler EILW measure presented in this article. The much more
complex methods they used do not readily lend themselves to making comparisons
across countries and over long periods, as the simple EILW so conveniently lets us
do. However, their much higher values of lifetime welfare suggest that the EILW
measure’s omission of the value of leisure most likely causes it still to underestimate
the true growth of individual welfare over the past two hundred years.
Another reason why EILW likely still understates individual welfare gains over
the past two hundred years is that it does not account for the fact that not only has the
average age increased, but the standard deviation of age at death has declined.
According to Lichtenberg, “People tend to live longer than they used to, and there
is also less uncertainty about the age of death. . . . If people are risk averse, they are
made better off by the reduction in the variance, as well as by the increase in the
mean, of the age of death” (1998, 2–3). This same point was also made some years
ago by the demographer Massimo Livi-Bacci (1997), who described the reductions in
12. For a review of the literature on estimating the value of life, see Viscusi 1993.
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 194
194
✦
H E N D R I K VA N D E N B E R G
infant mortality and the remedies for the many diseases that randomly attacked people of all ages as adding “order” to life and permitting people to more effectively plan
and organize their lives.
It is not difficult to imagine multiplicative adjustments for other economic and
social conditions that enhance or diminish the accuracy of real GDP per capita as a
measure of human welfare. For example, real GDP per capita might be multiplied by
a factor representing positive economic freedom, which, as suggested by Amartya Sen
(1999a, 1999b), improves as the economy increases its productive capacity. Certainly
the freedom to choose where one works and what one consumes is worth at least
some small percentage of national income. On the other hand, dictatorships, discrimination, threats to personal safety, and the lack of protection of personal property
diminish the real value of measured real GDP per capita. In this sense, Latin America’s
well-documented “lost decade” of the 1980s may not have been so “lost,” given that
life expectancy improved, access to education increased, and democracy replaced dictatorships in nearly all of the region’s countries.
Some extensions and refinements of EILW are worth pursuing in the future. For
example, the Disability Adjusted Life Expectancy (DALE) data just published by the
World Health Organization (WHO) are certainly a better means of measuring a person’s lifetime welfare than simple life expectancy (WHO 2000). These new estimates
of how long people can be expected to live in good health and to be able to enjoy fully
their material wealth, their education, and their lifestyles unfortunately exist only for
one year. Therefore, growth rates incorporating this measure cannot be calculated yet.
The issue of which measures best reflect human welfare remains to be settled, of
course. Any future improvements in the way we calculate GDP would also make
EILW a more accurate measure of lifetime welfare as well. However, the measures we
use to represent individual welfare will improve our understanding of individual welfare only if they reflect lifetime welfare.
Summary and Implications
The appropriate way to measure human welfare, in line with economists’ focus on
individual human welfare, is to quantify the economy’s capacity to provide the average individual with a lifetime of welfare. This focus demands that our measures of
economic performance reflect not just the average annual or some other point-intime level of welfare, but also how long the economy can keep the average person
alive to experience that level of welfare. Julian Simon’s statement in the epigraph to
this article suggests that the actual improvement in individual welfare has been much
greater than the customary measure of annual real GDP per capita shows. The longrunning debate about how to measure economic growth and human welfare really
has missed the mark by focusing on which annual flows or instantaneous stocks of
output and other “quality of life” variables best represent the growth of human welfare. It is surprising that development economists have never questioned using
THE INDEPENDENT REVIEW
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 195
ANNUAL REAL GDP AND THE GROWTH OF INDIVIDUAL WELFARE
✦
195
annual measures of human welfare, which completely ignore the “stupendous”
increase in longevity.
Perhaps the seemingly difficult task of compiling measures of lifetime welfare has
discouraged researchers from abandoning their annual and point-in-time stock measures. The difficulties in generating historical series for a large number of countries
using Nordhaus’s (2002) and Murphy and Topel’s (1999) methods seem to support
the impression that lifetime welfare measures are more theoretical than practical. This
impression, however, is wrong. A multiplicative measure, EILW—for which data are
readily available— captures lifetime welfare at least as well as GDP captures annual welfare. EILW has grown more rapidly than conventional measures such as real GDP per
capita. Given the “stupendous” increases in life expectancy, as well as the general
improvements in nearly all alternative measures of human welfare, lifetime measures of
welfare that reflect both the quality of life and the length of life confirm the general
impression given by Maddison’s estimates of real GDP per capita—namely, that individual welfare has grown enormously over the past two hundred years. In fact, lifetime
measures of welfare reveal that the improvement in individual welfare may have been
understated, not overstated, by the growth of annual real GDP per capita.
References
Anand, Sudhir, and Amartya Sen. 1999. The Income Component of the HDI-Alternative Formulations. Occasional Paper, United Nations Development Program. New York: Human
Development Report Office, United Nations.
Boskin, Michael J., Ellen R. Dulberger, Robert J. Gordon, Zvi Griliches, and Dale W. Jorgenson. 1996. Toward a More Accurate Measure of the Cost of Living. Final Report to the Senate Finance Committee, December.
———. 1998. Consumer Prices, the Consumer Price Index, and the Cost of Living. Journal of
Economic Perspectives 12, no. 1: 3–26.
Briscoe, Simon. 2001. Statistics May Not Measure Up in Quality of Life Test. Financial Times,
August 16.
Caplin, Andrew, and John Leahy. 2000. The Social Discount Rate. NBER Working Paper no.
7983, October. Cambridge, Mass.: NBER.
Federal Reserve Bank of Dallas.1999. The Right Stuff, America’s Move to Mass Customization:
1998 Annual Report. Dallas: Federal Reserve Bank of Dallas.
Fuess, Scott M.., Jr., and Hendrik Van den Berg. 1996. Does GNP Exaggerate Growth in
“Actual” Output? The Case of the United States. Review of Income and Wealth 42, no. 1:
35–48.
———. 1998. Does GDP Distort Mexico’s Economic Performance? Southern Economic Journal 64, no. 4: 973–86.
Kuznets, Simon. 1966. Modern Economic Growth. New Haven, Conn.: Yale University Press.
Lichtenberg, Frank L. 1998. Pharmaceutical Innovation, Mortality Reduction, and Economic
Growth. NBER Working Paper no. 6569, May. Cambridge, Mass.: NBER.
VOLUME VII, NUMBER 2, FALL 2002
TIR_7-2_c_Articles_asn 8/16/02 11:49 AM Page 196
196
✦
H E N D R I K VA N D E N B E R G
———. 2002. Sources of U.S. Longevity Increase, 1960–1997. NBER Working Paper no. 8755,
February. Cambridge, Mass.: NBER.
Livi-Bacci, Massimo. 1997. A Concise History of World Population. 2d ed. Oxford, U.K.: Blackwell.
Maddison, Angus. 1995. Monitoring the World Economy 1820–1992. Paris: Organization for
Economic Cooperation and Development.
———. 2001. The World Economy: A Millennial Perspective. Paris: Organization for Economic
Cooperation and Development.
Meier, Gerald M. 1976. Indicators of International Development. In Leading Issues in Economic
Development, 3rd ed., edited by Gerald M. Meier, 12–18. New York: Oxford University Press.
Murphy, Kevin M., and Robert Topel. 1999. The Economic Value of Medical Research. Paper
presented at the Conference on the Economic Value of Americans’ Investment in Medical Research, sponsored by the Lasker Charitable Trust, December 1999.
———. 2000. Medical Research: What Is It Worth? Milken Institute Review 2, no. 1: 23–30.
Nordhaus, William D. 2002. The Health of Nations: The Contribution of Improved Health to
Living Standards. NBER Working Paper no. 8818, March. Cambridge, Mass.: NBER.
North, Douglass C., and Robert Paul Thomas. 1973. The Rise of the Western World: A New Economic History. Cambridge: Cambridge University Press.
Pritchett, Lant. 1997. Divergence, Big Time. Journal of Economic Perspectives 11, no.3: 3–17.
Robinson, John P. 1995. Trends in Free Time. In The State of Humanity, edited by Julian L.
Simon, 224–30. Oxford, U.K.: Blackwell.
Sen, Amartya. 1999a. Development as Freedom. New York: Alfred A. Knopf.
——. 1999b. The Possibility of Social Choice. American Economic Review 89, no. 3: 349–78.
Simon, Julian L., 1995a. Introduction to The State of Humanity, edited by Julian L. Simon,
1–28. Oxford, U.K.: Blackwell.
———, ed. 1995b. The State of Humanity. Oxford, U.K.: Blackwell.
United Nations Development Program (UNDP). 1999. Human Development Report 1999.
New York: UNDP.
Usher, Dan. 1973. An Imputation to the Measure of Economic Growth for Changes in Life
Expectancy. The Measurement of Economic and Social Performance, edited by Milton
Moss, 193–223. New York: Columbia University Press.
———. 1980. The Measurement of Economic Growth. New York: Columbia University Press.
Van den Berg, Hendrik. 2001. Economic Growth and Development. New York: McGraw-Hill.
Viscusi, W. Kip. 1993. The Value of Risks to Life and Health. Journal of Economic Literature
31, no. 4: 1912–46.
Weitzman, Martin L. 1999. Pricing the Limits to Growth from Minerals Depletion. Quarterly
Journal of Economics 114, no. 2: 691–706.
World Health Organization. 2000. WHO Issues New Healthy Life Expectancy Rankings.
WHO Press Release, June 4. Figures available on the WHO Web site: www.who.org.
Acknowledgments: The author would like to acknowledge the reliable research assistance of Ying Zhou and
the valuable suggestions by two anonymous referees. The author is solely responsible for any remaining errors.
THE INDEPENDENT REVIEW